Project Proposal

SINTEF collaborate closely with Norwegian University of Science and Technology (NTNU), and University of Oslo (UiO), both in research projects and in supervising students on Master and PhD levels. The department for Networked Systems and Software (NSS) in SINTEF ICT takes part in this project. SINTEF ICT consists of about 300 employees, and 9 research departments. NSS provides research-based expertise in three main areas: model-driven development, quality and security technology, and user-centred development. NSS has extensive experience in EU projects, both as the coordinator and participant such as in FP6 projects: ATHENA, INTEROP, MAPPER, MODELPLEX, MODELWARE, SODIUM and SWING, and in FP7 projects: DiVA, IRMOS, SHAPE, REMICS and NEFFICS. NSS has developed various meta-models and languages, methods and tools for model-driven system engineering and participates in OMG MDA standardisation. The recent activities have been the standardisation process of Common Variability Language (CVL), Service-oriented architecture Modeling Language (SoaML) and work undergoing in the REMICS project to define modeling abstractions for IaaS.

Based on this experience, we plan to investigate how cloud computing and models can be efficiently coupled to support the end-user while modeling and deploying software. This project proposal, called Mod4Cloud leverage the Amazon infrastructure to experiment in three main research directions:

Experimentations

Data Mediation

Our mediation engine complements existing ontology-based approaches to identify mappings data models, with a special focus on object-oriented models. By combining graph exploration techniques and fuzzy logic, the engine identifies recommendations to support the user during the mediation process (e.g., “element X matches element Y with 96% of accuracy”). The user might then refine the suggested mediation, which is later compiled into a web service actually bridging between two existing services.

Mod4Cloud focuses on:

  1. improving the efficiency of the engine thanks to an Elastic Map Reduce implementation,
  2. providing automated techniques to host the generated services in Amazon EC2, and
  3. conducting experience using large-scale data models stored in Amazon S3.

Application Modeling

Our Cloud modeling language (CloudML) provides the user with cloud-specific concepts to be used to model software, including elastic capability, geographic zone, failure recovery, and data protection. The language acts as an intermediate pivot between users and clouds. A compiler is used to project this language into concrete cloud instances, and support the deployment.

Mod4Cloud intensively relies on Amazon EC2, and more specifically the Amazon Cloud Formation language. It will allow users to bind their code with EC2 instances, and the compiler will generate both:

  1. the associated Cloud Formation template to support provisioning, and
  2. the associated deployment script required to properly deploy the modeled application on the provisioned instances.

Internet of Things & Sensor Networks

Based on our experience in environmental monitoring through sensor networks, we are currently developing SensApp, a portal for sensor networks and Internet of Things (IoT). SensApp acts as a RESTful platform where heterogeneous sensors can push data. The extremely large amount of data produced by sensors is stored in SensApp, and can be exploited by the end user and visualized in different widgets. We also provide mechanisms to create virtual sensors, that is, the composition of existing elementary sensors into more complex one. SensApp relies on ThingML, a languaoe usadto model IoT systems. ThingML support the generation of platform-specific code (e.g., Arduino) to efficiently deploy code on sensors.

In this context, Mod4Cloud relies on:

  1. Amazon SimpleDB to store the data sets sent to SensApp,
  2. Amazon Elastic Map Reduce to process these data sets, and
  3. Amazon Simple Queue Service to accurately implement asynchronous communication to support the implementation of virtual sensors.

Amazon Cloud Usage Requirements

Mod4Cloud will cover the different aspects of the Amazon cloud offer.

  • The data mediation axis investigates the scalability of the recommendation engine. Thus, it will highly rely on Map/Reduce algorithms to properly support the mediation. In this context, Mod4Cloud will intensively exploit the computational resource from the Amazon cloud.
  • The application modeling axis focuses on instance provisioning, filling the gap between the level of abstraction of the user domain and the technical reality of clouds. Here, Mod4Cloud will mainly perform experiments on automatic deployment to Amazon EC2, provisioning a large number of instances for a short period of time.
  • Finally, the SensApp platform relies on the scalable storage of data from sensors. Thus, Mod4Cloud will rely on data storage capabilities, to store large data sets for a long period of time.

In all the above axes we need access to the global computing infrastructure and storage capacity of Amazon Web Services. The project will also involve master students and has a research / education perspective.

Work Plan

Mod4Cloud acts as a two-phases project. The first phase (year 1) consists in conducting proof of concepts experiments with Amazon cloud tools to properly connect our internal tools with the Amazon technology. The second phase (year 2) will use the remaining credits to support large-scale experiments based on our 3 scenarios. Results will be published in renowned conferences and journals, acknowledging the Amazon support.

Involved Participants

  • Dr Arne-Jørgen Berre, Associate Professor II (University of Oslo), Senior Research Scientist (SINTEF ICT)
    • Arne-Jørgen is chief research scientist at SINTEF ICT. He received his Ph.D. from the Norwegian University of Science and Technology in 1993 and is Associate Professor II at the University of Oslo Institute for Informatics since 1996. He has served as technical manager and scientific coordinator for a number of IST projects, including the Eureka Software Factory as early as 1988-1991, COMPASS, OBOE, COMBINE in the EU 5th framework and ATHENA, INTEROP, and SHAPE. He is involved in interoperability and model-driven standardization activities in OMG and OASIS.
  • Dr Franck Chauvel, Research Scientist (SINTEF ICT)
    • Franck is currently working in SINTEF ICT as a research scientist. Formerly, he joined the Software Engineering Institute of the Peking University as a postdoctoral fellow, working on the convergence of Software Architecture, Middleware and Models at runtime towards effective self-adaptive systems. He holds a Ph.D degree and a Master degree in Computer Science from University of South Brittany. He has contributed to various national and an international research projects such as EU FP7 CONNECT or the EU FP7 REMICS.
  • Dr Franck Fleurey, Research Scientist, (SINTEF ICT)
    • Franck is a research scientist at SINTEF ICT. He received a Ph.D. degree in Computer Science from the University of Rennes 1, France in 2006, for a dissertation on model-driven development techniques. Part of his doctorate work has been to propose the Kermeta language. He is the authors of over 45 peer-reviewed scientific publications on software engineering. He joined SINTEF in 2008 and works on practically applying model-based software engineering technique to embedded systems, product lines and adaptive systems. He has experience from working in a number of EU research project and has been a work package leader in the DiVA FP7 STREP project.
  • Dr Parastoo Mohagheghi, Associate Professor II, Norwegian University of Science and Technology (NTNU)
    • Parastoo received her Ph.D. in Information and Communication Technology in 2004 from the Norwegian University of Science and Technology where she is also Associate Professor II. She has published articles in conferences and journals in several software engineering areas and serves as a reviewer for international journals such as Empirical Software Engineering journal, SoSym and Information and Software Technology journal, and has also served as organiser of several international events. She has experience from EU projects such as participating in SHAPE, leading model engineering work package in MODELPLEX, and is currently the coordinator of the REMICS project (http://www.remics.eu).
  • Dr Brice Morin, Research Scientist (SINTEF ICT)
    • Brice is a research scientist at SINTEF ICT. He received a Ph.D. degree in Computer Science from the University of Rennes 1, France in 2010 for a thesis entitled “Leveraging Models from Design-time to Runtime to Support Dynamic Variability”, funded by the DiVA European project, where he pioneered the concept of models@runtime, in close collaboration with DiVA partners. He served as Work Package leader for several EU projects including FP7 DiVA and FP7 REMICS. He is interested in applying Model-Driven Engineering and models@runtime techniques to adaptive systems, IoT-based or cloud- based systems.
  • Dr Sébastien Mosser (Principal Investigator), Research Scientist (SINTEF ICT)
    • Sébastien is a research scientist at SINTEF ICT. Previously, Sébastien was postdoctoral fellow at INRIA Lille-Nord Europe. He holds an Engineering degree (2007) in Computer Science from the Polytech’Nice-Sophia school of engineering, and a PhD degree (2010) in Computer Science from the University of Nice. He had served in the organizing committees of several workshops and events, and served as a reviewer for high-level conferences and journals, including CBSE, SOCA, TOOLS, ECMFA & SoSym. He is also a founding member of the ”Computer Science Night” (since 2007), an international event that federates up 23 universities and involves up to 1,300 students once a year into a night-long programming contest.
  • Dr Arnor Solberg, Group leader, Senior Research Scientist (SINTEF IKT)
    • Arnor is a senior research scientist at SINTEF ICT and the research manager of the Model-Driven Software Development group. He is an expert on software architectures and software engineering practices. He received his PhD from the University of Oslo. His research interests include model-driven software engineering, aspect oriented modeling and adaptive systems. He has a long track record as a project manager and technical manager for several national and international/EU research projects, in particular he was the technical manager of the DiVA FP7 project. He is member of program committees of relevant international conferences and workshops such as the MODELS conference and the annual models@runtime workshop (Satelite event of MODELS).

References

  • [2011] Sébastien Mosser, Eirick Brandtzæg, Parastoo Mohagheghi. “Cloud–Computing: from Revolution to Evolution” (extended abstract) in BElgian-NEtherlands software eVOLution seminar (BENEVOL'11), VUB, Brussels, Belgium, 8-9 december 2011 (PDF) (Slides)
  • [2011] Franck Fleurey, Brice Morin, Arnor Solberg and Olivier Barais. MDE to Manage Communications with and between Resource-Constrained Systems. In proceedings of 14th International Conference on Model Driven Engineering Languages and Systems, MODELS 2011, Wellington, New Zealand, October 16-21, 2011. (PDF)
  • [2011] Franck Fleurey, Brice Morin, and Arnor Solberg. A Model-Driven Approach to Develop Adaptive Firmwares. In SEAMS’11: 6th International Symposium on Software Engineering for Adaptive and Self-Managing Systems. Waikiki, Honolulu, Hawaii, USA, May 2011. (PDF)
  • [2011] Roman D., X. Gao, and A.-J. Berre. Demonstration: Sensapp ⎯ An Application Development Platform for OGC-based Sensor Services. Proceedings of the 4th International Workshop on Semantic Sensor Networks 2011 (SSN11), collocated with ISWC 2011, October 23-27, Bonn, Germany. (PDF)

Documents

projects/proposal/mod4cloud.txt · Last modified: 2012/02/10 00:00 (external edit)